Already, the president’s team has announced the cancellation of dozens of contracts through the Institute for Education Sciences, the Department of Education’s research arm. Over 1,300 of the department’s employees, amounting to roughly half of its workforce, have been terminated, casting doubt on whether key functions like national testing initiatives can carry on without interruption. And the future of dedicated learning hubs, including one credited with triggering a breakthrough in reading instructions, is in serious doubt. 

The wave of cuts and firings was the unspoken agenda item at the 50th annual convening of the , one of the most prominent professional organizations for education researchers. In mid-March, amid three days of panels and paper sessions touching on every conceivable topic in K–12 schooling, hundreds of academic economists, education activists, graduate students, and district staffers exchanged concerns about the future of public insight into schools.

Ironically, those worries emerged just as AEFP unveiled a critical new tool: its of education policy research, gathering and distilling the findings of thousands of studies. Its 50 chapters address a bevy of questions ranging from preschool to higher education, including the makeup of local school boards, performance of charters and school vouchers, teacher preparation programs, the effects of education spending, and more. The Association hopes the extensive and growing site, an update of previous printed versions, can provide educators and lawmakers alike with something akin to a Wikipedia for research.

Leading the effort is Tulane University economist Douglas Harris, a veteran researcher who also heads the and the . An expert on charter schooling, Harris has been one of America’s most productive scholars studying how district, state, and national policies shape what kids learn — and addressing some of the most contested questions in the field, including whether school choice actually improves the delivery of education.

In a conversation with ����Ӱ��’s Kevin Mahnken at the conference in Washington, Harris talked about the origins of the Live Handbook project, how its creators intend to ward off ideological bias, and why IES and other federal research efforts are irreplaceable supports in the U.S. education infrastructure.

“In some sense,” he said, “the Live Handbook is a monument to IES, at a time when IES is being knocked over.” 

What’s the purpose of this project?

The idea is to make research more useful and make researchers more useful. One of our purposes is to just get research summarized and discussed in a way that’s actually accessible to a broad audience, and another is to connect researchers to policymakers and journalists. If you’re looking for an expert on an issue, you can find their names in these articles, click on them to get their information, and just email that person. 

We’re hoping users create these networks of expertise and connect them with people who need that expertise. 

I have to say, it sounds like you’re trying to put education journalists out of work.

So I think this will be very useful for journalists, who will be able to find something much easier and link to it in their stories. And hopefully, when somebody is searching for a topic, the handbook will come up as the top result, which will make it build further. The more reach it has, the more people will want to write for it, and the more existing authors will want to update it — which is another important part of the document. It’s not just static, it’ll be updated every year, and all the authors will be expected to continue working on it. If they decide they don’t want to do that, they’re going to hand it back to us, and we can turn over the authorship to somebody else.

That kind of arrangement is actually unusual from the standpoint of intellectual property. We weren’t sure how that was going to work at the beginning, or if it was even legal to do it that way. But it turns out that, as long as everyone is clear about it, you can write the agreement that way. Part of the motivation for this project was to marry the traditional handbook with Wikipedia, but with Wikipedia, there’s no issue with authorship.

I didn’t realize you took inspiration from Wikipedia.

Well, there had been some talk of doing another handbook, which we’d been doing just about every decade. They were all about 700 pages long, there’d be 30 or 40 chapters written both for and by academics. We’d mostly use them as syllabi and readings for education policy classes, but the only real audience was other AEFP members. 

The other inspiration emerged from at Education Research Alliance-New Orleans. We had something like 40 studies on our website, and was like an integrated summary. All this evidence was just for the relatively narrow topic of school reform in New Orleans. 

The question we were asking ourselves, and which you should always ask if you’re writing something, was who our intended audience was. What we realized that we didn’t have to choose between researchers and policymakers. The beginning of each entry looks like a policy brief, and non-researchers will probably stop when they get to the last key finding. But if you want more detail, you just click on that finding, which takes you to the longer discussion that would have been included in a printed version of the handbook. If you want even more than that, you can click on the endnotes, which will hyperlink you to the underlying studies themselves. 

So you’re serving everybody: At the top, policymakers are your main audience, but by the time you get to the bottom, the researchers and experts in the field can dig in.

Up to this point, would you say that education research has been effectively communicated to the public, and that it has informed how politicians create policy and oversee schools?

Uh, hard no. [Laughs] We have not done a good job with those things.

There have certainly been moves toward that. There’s something called the Research Practice Partnership movement, which is supposed to develop genuine partnerships between the research and policy worlds, and it’s great. But they’re really hard to create and sustain, and they tend to be very localized. We wanted to do something that had broader research.

We’ve also got the What Works Clearinghouse, which is federally funded. If you look at those releases, though, they never realized their potential. They were too slow, they were written by committee, not very readable. All of this was aiming in the right direction, but not hitting the target well. There was clearly a hole there that we’re now trying to fill. 

As you mentioned, the is a federal resource that’s only a few decades old — although, given reports that its funding has been cut, it may not get much older. Is the need for a live handbook related to the fact that the social science around education outcomes doesn’t go back very far?

The federal government certainly led the way in moving toward evidence- based policy. And everything I just mentioned emerged out of that orientation, which was mandated by law around the time of No Child Left Behind and is still in effect today. 

There is also a natural demand, in the sense that people want to do the right thing. They want to make their K–12 schools and colleges better, and they want advice. But advice is usually pretty ad hoc; it depends on who’s in your network, who’s got a friend in a school nearby, and what they’re hearing. There will always be a place for that, but having actual evidence at the root of those conversations has a lot of potential to improve things.

Can you think of an area of research where the evidence has managed to break out of academic discourse and influence the public? 

Research doesn’t drive most conversations about policy and practice, but it can have influence at the margins and create new ideas. The science of reading is the example that comes to mind immediately. Russ Whitehurst, who became the first director of IES, is a psychologist, and he was the one who the reading research. Almost all the underlying evidence for that is IES-funded research, which is noteworthy under the current circumstances.

Another example would be class-size reduction. There was a lot of interest in that for a while, until it became clear that, while it works pretty well, it’s also . The school funding debates, and whether money matters in student learning, would be another case.

I think people can get their arms around class sizes and budgets being important issues in schooling. But even for someone like me, who has experience consulting research, it can be very difficult to weigh the evidence that various experts marshal on questions like teacher evaluation or early childhood education.

That’s the hole we want to fill, right there. We want you to do that google search and come to what we’re doing because we have answers to those questions.

We’ve got 50 chapters in this first round, and the plan is to update each of those next year while adding another 25. Part of it depends on funding, and growth is very time-intensive. We have to do all the things that a publisher does, putting it on the website and creating PDF versions and all that. But the idea is to grow the handbook so that it becomes comprehensive, both in terms of covering every part of education — early childhood through higher education — and also trying to cover all the key policy areas.

How comprehensive is comprehensive? There are really old, but foundational studies, like on segregation and achievement gaps, which were conducted when methods were much more crude. Are you trying to include that kind of evidence? 

What we’re doing is an awful lot without also trying to write the history of education research. The most recent research is obviously more relevant because context matters. The world is changing around us, and that affects education. So we want more recent studies.

Another important thing to remember is that recent studies tend to be methodologically better — again, partly because of IES and the principles and demands that IES has placed upon us. We’re telling authors to focus on the most recent and best studies, because those are going to be more useful to the field.

In some areas, like school finance, the debate among leading researchers still burns very hot. I’m sure it’s difficult to arrive at anything like a consensus, so are you just trying to represent the state of play?

It’s a challenge. From the beginning, we didn’t expect that we would release these chapters and people would say, “Oh, you’re exactly right!” Sometimes they’ll say, “Wait a second, I don’t agree with that.”

I wrote the charter school section, and we showed it to a group of policymakers and practitioners who are advising us. We brought them into the design process to make it more useful to them, and when I was doing a show-and-tell a couple of weeks ago, somebody said, “I’m not sure I agree about your point on charter schools!”

We knew that was going to happen. So we’ve advised the authors to present different sides of the debate — the major positions that are pretty widely known — and, if a key finding falls clearly on one side, then at least address what the other side of the argument is. We’re being really clear about the evidence base, and I can’t just say, “Such-and-such is true of charter schools because I feel like it.” It has to be because one group of charter studies is stronger than another group of studies, or there’s something really unusual about the context of some studies that make them less convincing. Basically, there has to be a reason that addresses the different sides of arguments.

What structures have you put in place to prevent a kind of ideological drift?

Our editorial board helps with that. It’s a very wide-ranging group where you have some people who are seen as more on the left, and some who are more on the right. There are people from different disciplines. We’ve encouraged authors to include studies from outside their disciplines; quantitative people should include qualitative work, and vice-versa.

The board is there to enforce that and make sure we were getting the right range of perspectives, so that when there are disagreements, someone can say, “Hold on a second, you’re missing something.” We’re not going to be perfect in the first round, but the process is set up to be updated and receive feedback. If you’re reading a piece and have a question, or you want to debate some point, you can click the feedback link and explain it. We’ll send all that information to authors, and once a year, they’ll be expected to go through those comments. If we see significant issues with a piece, we’ll push them to update it.

And I imagine the various writers and editors can weigh in on other entries as well. 

Here’s the way we handle these discussions: I wrote the charter schools chapter, which was edited by . He’s the one pressing me on the evidence there, but I’m the editor for his section on school vouchers. We view those topics a little bit differently, but that’s a way we enforce objectivity. We recognize that everybody’s subject to that sort of bias.

That’s an interesting pairing. From my perspective, the public debate around charter schools — which has been extremely contentious in the past — has become somewhat quiescent, while the voucher issue has just roared into prominence over the last few years.

There are a lot of studies in play on vouchers, so Pat will probably have to update his chapter next year, and every year after that. Much of the research in that area is old and based on the city-based voucher programs in places like Milwaukee or Washington. Then you had the four states where we could study statewide voucher programs, which are probably the most relevant to the current discussion. And we’ll also be including three or four national studies that we’ve got going at the REACH Center, which I lead.

Part of the problem with the way the new voucher programs are set up is that, in a sense, they’re designed not to be studied. There’s no state testing requirement, so we don’t have test-based outcomes, and you’re confined in what you can study. Still, there will be a lot of interest in that topic.

What do you make of the cuts to federally supported research that have been announced over the last month?

It’s a very big deal. If you look at the endnotes for this handbook, probably half of them have a basis in IES. Either the studies themselves were funded by IES, or they’re using IES-funded data sets, or they’re written by researchers who were in the IES pre-doc or post-doc programs. There was a whole set of training programs that were designed to develop the next generation of scholars. So in some sense, the Live Handbook is a monument to IES, at a time when IES is being knocked over. 

It’s very worrisome, and I think we’re at a really uncertain time. I don’t think IES is going to go away. But we don’t really know what direction it’s headed in or how long it will take to get there, given that they just fired essentially everybody. The best-case scenario is that they hire a new director who’s allied with the administration and who has sympathy and a desire to build it back up. There’s no question that there are ways the institute could be made better, but there are also a lot of things you’d want to keep about the old structure. 

It’s good to have decisions made by people who are researchers and know the field. It’s good to have policymakers involved in decisions about what gets funded, which has been true for a long time. Should the research process be faster? Sure, we could find ways to do that. So it’s possible that IES comes out better at the end of this. But will it? It’s a huge question right now. 

There’s obviously a lot of concern at a conference like this, where people have seen IES as the root of so much of the work we’re doing.

Virtually every researcher I’ve spoken to has said something similar. People will generally concede that improvements can be made, but where the process calls for a scalpel, DOGE is using a dump truck.

I think that’s right. In the amount of time they had, they couldn’t have possibly learned what grants or contracts should be kept. If you’re trying to do it based on reason, there’s no way to do it in a matter of weeks. It’s been very arbitrary, just searching for keywords and things like that. It’s no way to fix anything, it’s a way to knock things down.

Something people don’t realize is how long it took to build IES to begin with, and to gain support for it. It started, I believe, back in 2001, and it took a long time to build up the staff and the expertise. Especially in terms of data collection, it’s just underestimated how much expertise goes into what IES does. All these contracts with Mathematica and AIR depend on those organizations’ very significant internal capacity in areas like getting schools and students to respond to surveys. It doesn’t just happen. There’s so much expertise that goes into those tasks, which you’ve now destroyed.

Even if they succeed in making things better in other ways, that’s going to make it much harder to build back the things they should want to keep. It’ll be like a wave pushing against them.

Mark Schneider, the IES director under both Presidents Trump and Biden, told me that the original intention was for the institute to grow much more substantially than it has, until it more closely resembled a $40 billion agency like the NIH. Even though that hasn’t happened, it has punched above its weight in expanding the knowledge base about schools.

Oh, absolutely. When you think about what a good organization of any kind spends on R&D, it’s a much greater proportion than the IES budget relative to total education spending. IES has about a $1 billion budget, and the United States spends something like $700 billion per year on education. So that’s less than .2 percent. It’s less than any standard you could come up with.

It’s always been underfunded, and they use those resources well. Collecting data, for example, creates so many positive spillover effects because once you’ve collected it, anyone can use it. The pre-doc and post-doc programs are really important for producing people who can work at school districts and state agencies, which need professionals who are really trained in research. That may go away too. 

One of the things that doesn’t get enough attention is that the federal government was very involved in creating the state longitudinal data systems, which have played an enormous role in just about every area of policy research. The federal government gives money to the states to create those systems and make them available, and they allow us to link schools and programs to student outcomes. Without that, you’ve got nothing. 

You mentioned that IES was instrumental in generating research on the science of reading, too.

That’s probably the best example of the organizational influence. It’s not so much about the data they were collecting, but it was related to the projects they were funding. It’s also a good example of something Republicans support. They’re all about the science of reading, but what’s happening is that they’re basically undercutting the next science of reading. 

There might be some research studies that don’t seem very useful, but in a way, that’s the point of research. You don’t know what’s useful until you actually do it. We don’t know what the next science of reading is going to be. Hopefully, the Live Handbook will help find it, but we’d find it faster with IES underpinning the research that will get us there. 

The report, released Tuesday morning by Stanford University’s Center for Research on Education Outcomes, provides perhaps the most thorough perspective available of the landscape of charter schooling, which has grown significantly in recent years.

Macke Raymond, CREDO’s founder and director, said that the report sketched a picture of continuous improvement for the charter sector over the last 15 years. The center’s first national analysis, issued in 2009, showed charters under-performing traditional schools in both core subjects; in a 2013 follow-up, they slightly bested traditional schools in English while still lagging in math. That movement represents a modest silver lining for American education, she said, after a prolonged period during which learning — as measured by standardized tests like the National Assessment of Educational Progress — largely stagnated even before the pandemic. 

“When you compare [our findings with] the results of NAEP — which, over an equivalent period, have completely flatlined — what you’re looking at is really the only story in U.S. education policy where we’ve been able to create a set of conditions such that schools actually do get better,” Raymond argued.

The new study focuses on charter school performance in 29 states, as well as Washington, D.C., and New York City, incorporating standardized test scores between 2015 and 2019. All told, over 80 percent of tested public school students were included in CREDO’s data set. More than 1.8 million charter students were each paired with a “virtual twin” (i.e., a nearby pupil possessing similar demographic traits and prior test scores) enrolled at the district school that the charter student otherwise would have attended.

The research team calculated that charter school students gained the equivalent of an additional 16 days of learning (based on a traditional 180-day school calendar) in English compared with similar kids at district schools. Their six-day edge in math was smaller, though still considered statistically significant.

But even those averages, comprising millions of student measurements across the country, contain significant variation. Black students attending charter schools gained 35 days of growth in reading and 29 days in math — as if they’d attended school for an extra 1.5 months over a single school year. Hispanics enjoyed 30 extra days of reading and 19 in math. By comparison, white and multiracial students lost the equivalent 24 days of annual math learning in charter schools. 

Smaller sub-groups experienced similar divergences. Poor students saw much higher gains in charters than in traditional public schools (23 extra days of reading growth, 17 extra days in math), as did English learners (six extra days of reading, eight in math); students with overlapping designations (such as both African American and low-income, or both Hispanic and English learner), also made considerable strides

By contrast, special education students were seriously stymied, losing 13 days of reading growth and 14 days of math at charter schools relative to kids receiving special education outside of charters. Raymond called that inequity one of the few sore spots revealed by the study, adding that charter schools should be “taken to task” for the collective failure.

“With the exception of very few charter schools that specialize in particular kinds of special education, the sector has basically thrown up their hands and said, ‘This isn’t our job,’” she said.

Even among charters, some types tend to yield better results than others. Specifically, those grouped within a charter management organization (CMO) — a network, either non- or for-profit, that operates multiple schools, such as the well-known KIPP or Success Academy organizations — provide 27 extra days of instruction in reading, and 23 extra days in math, than traditional schools. Stand-alone charters, which encompass roughly two-thirds of all charter schools, generate 10 extra days of reading growth and negative-three days of growth in math.

Douglas Harris, an economics professor at Tulane University the impact of charter schools on surrounding public school districts, said that the results of the CREDO report largely dovetailed with those of in New Orleans and elsewhere. He also said that the especially impressive findings from CMO-affiliated schools were somewhat predictable given that many cities and states only consider top-performing charter schools as candidates for replication.

“Some of this is kind of mechanical — not in a bad way, it’s just how the sector operates. If you’re a stand-alone, and you do well, you can open another school,” Harris said. “Then you become a CMO, and they’re better because they were selected to build on their own success. That’s a positive aspect of the charter model.”

Even more distinctive was the dividing line between what might be deemed “traditional” charters and those offering instruction virtually, which had already earned an ugly reputation for low academic quality even before the pandemic began. The popularity of the virtual charter sector has grown substantially since the emergence of COVID — by the Network for Public Education found that fully or mostly online programs enrolled 13 percent of all charter students during the 2020–21 school year — even as they delivered a staggering 124 fewer days of math growth than traditional public schools, along with 58 fewer days of growth in English.

If virtual initiatives were excluded from the national sample, the average charter school advantage would jump from 16 extra days of reading instruction to 21, and from 6 extra days of math instruction to 14. 

Martin West, the academic dean at the Harvard Graduate School of Education, called the report “easily the most comprehensive analysis of charter school performance to date” and echoed concerns about the performance of virtual charter schools.

“The results continue to raise questions about the regulatory environment for virtual charter schools, whose results drag down the overall performance of the broader sector,” West said. “These schools may provide an essential option for students for whom in-person learning truly isn’t possible, but state policymakers should look carefully at who is attending these schools and how well they are being served.”

An additional state-by-state analysis showed that individual jurisdictions have built particularly effective charter school sectors. Across New York State, charter students receive the equivalent of 75 extra days of growth in reading, and 73 extra days in math, compared with demographically similar students at district schools. Massachusetts (41 extra days in both subjects), Maryland (37 extra days in both subjects), Tennessee (34 extra days of reading and 39 in math), and Rhode Island (90 extra days of reading and 88 in math) offered similarly impressive statewide results. Charter school students only experienced significantly weaker reading growth in one state, Oregon.

An additional lesson came with respect to new charter entrants versus existing options. New schools opened by existing CMOs tended to outpace their district competitors, but also to be out-performed themselves by older schools within their own CMO.

“The new schools that have come in since the second study are strong, but they’re not as strong,” Raymond observed. “So it’s not that new schools are coming in and kicking butt and dragging the sector along with them. It’s that, over this period, individual schools around the country are making incremental changes that lead to this trajectory of upward performance.”

Public charter schools may have lost some of the luster they enjoyed with centrist Democrats in Washington, D.C., a decade or two ago, but a top Biden administration education official this week sought to reassure the sector that it enjoys broad support on both sides of the aisle.

“I do not believe that the bottom has fallen out from under the bipartisan coalition for public charter schools,” said Roberto Rodriguez, assistant secretary for planning, evaluation, and policy development at the U.S. Department of Education. “I think if that were the case, you would see the funding completely deteriorating from this program. And in fact, you’re not seeing that.”

The Biden administration has faced harsh criticism for its stance on its $440 million , a key federal grant that more than half of charter schools rely upon. This comes as centrist Democrats, once the sector’s biggest backers, have sought political support from teachers’ unions, which for decades have forcefully opposed charters.

During the 2020 presidential campaign, then-candidate Joe Biden admitted, “I’m not a charter school fan.”

But on Wednesday during a panel discussion at Washington, D.C.’s Brookings Institution, Rodriguez adopted a softer posture.

“We support high-quality public schools for all kids, including high-quality public charter schools,” he told Brookings Nonresident Senior Fellow Doug Harris, the panel’s moderator. “Our budget stands behind that. The work we’re doing stands behind that. The rulemaking that we’ve proposed is not an effort to tear down the charter school sector. In fact, it is an effort to further promote that objective.”

But the administration has warned that more than one in seven charter schools funded by the grant either never opened or shut down before their grant period ended, in effect wasting an in taxpayer funding. In response, last year it proposed new regulations that critics said amounted to a new “war” on charter schools.

The originally proposed rule for applicants required them to prove their schools met “unmet demand” in existing public schools — a requirement that charter advocates said ignored a bigger problem in district schools: poor quality.

The department also said applicants had to collaborate with “at least one traditional public school or traditional school district,” in effect giving districts a veto over their plans, according to charter advocates.

A third requirement said charter schools had to show they wouldn’t worsen district desegregation efforts or increase racial or socio-economic segregation or isolation in schools.

Taken together, , the draft requirements were “tailor-made to ensure that the most successful charter schools won’t be replicated or expanded.”

The education department received 26,550 comments on the proposed regulations, andangry charter school parents the White House in May to protest Biden’s stance on funding regulations.

eventually admitted that the final rules, issued in August, were less harmful but “not without impact” on future growth of the sector. Among the concerns: a shortened window for submitting applications.

Two groups , saying, among other things, that the department lacked authority to impose new criteria on the grants, which Congress approved as part of a massive spending bill in December. It level-funded the charter grant for the . 

Harris, who has long studied the sector, noted that recent campaign rhetoric “has been different from what the actions have been in the administration,” with more public-facing skepticism from lawmakers about charters than “what’s happening in the nuts and bolts of committee rooms.” He asked the panel if they see the coalition for charters “fracturing” on the ground, especially among centrist Democrats.

Shavar Jeffries, CEO of the KIPP Foundation, which trains educators for the network’s 280 schools, observed that even in the movement’s “halcyon heydays,” charters were simultaneously “contentious among a variety of different constituencies” and the beneficiaries of significant bipartisan support. That continues today, he said.

“I do think there’s a kind of false idea [that] people are moving away from the issue in ways that [are] maybe inconsistent with what we’ve seen in the past,” he said.

But Jeffries said opponents of the Biden regulations had a point about not wanting to collaborate with districts, since some district officials are “not interested in the practices we’re trying to share.” He added, “You can take a horse to water, but you can’t take it much further than that [if] people aren’t interested.”

In a few instances, Jeffries said, opponents “are actually acting aggressively to undermine the capacity for public charter schools to exist.” He recalled local superintendents who were not only opposed to KIPP practices, but “sadly, in some instances…didn’t even want us to be here. So the idea that we’re going to obtain their support is obviously not going to happen.”

He also said the requirement that charter schools not worsen segregation can, in some cases, amount to a requirement that schools serving Black and Latino students essentially find white students in the suburbs.

Charter schools serve more than 3 million students, recent research shows, about two-thirds of them Black or Hispanic and most low-income. 

The Brookings panel also included from another panelist, Katrina Bulkley of Montclair State University, who led a team that found charter school authorizers are a key but little-studied aspect of the charter school world.

While some authorizers say equity is key to their mission, they found, others focus on choice or “market logic.” And they found that authorizers that prioritized equity received applications from schools that also prioritized equity. “This really suggests to us that those beliefs and the practices of authorizers are shaping what applicants are submitting,” Bulkley said.

Correction: An earlier version of this story contained an incorrect funding amount for the federal Charter Schools Program.

What made so many K-12 schools stick with remote learning to begin the 2020-21 school year, even as others reopened their doors to in-person or hybrid instruction?

According to a slew of research that emerged last year, much of the answer boils down to simple politics. Multiple studies from political scientists at Michigan State University, Boston College, and the Brookings Institution suggested that school reopening decisions were significantly more correlated with local political affiliation — as measured by the results of the 2016 election, or sometimes by strength of teachers’ unions — than the prevalence of COVID-19. Like so many other events in American life, our responses to the pandemic were heavily governed by how we vote.

But a new analysis complicates that picture somewhat. The paper, released on Monday by Tulane University’s National Center for Research on Education Access and Choice, confirms that politics helped determine how local authorities responded to the coronavirus threat in schools. But reopening approaches were also closely correlated with community demographics, and health conditions played a role as well. What’s more, given the interdependent relationships between all of those variables, it’s extremely challenging to isolate just one as being the most influential on policy makers’ decisions last fall.

Douglas Harris, an economics professor at Tulane and one of the report’s co-authors, said in an interview that the mix of potential causes yielded “probably the least amount of certainty about what the actual conclusions are of any study I’ve ever done.”

“In this case, that’s part of the point,” Harris said. “What we thought was certain is actually a little more uncertain, and given the political polarization we’re in right now, we don’t want to overstate the role of politics here.”

Using data on over 1,100 school district reopening decisions from Burbio.com, Harris and Tulane research fellow Daniel Oliver examined which stayed remote or switched to in-person learning both last fall and in the spring of 2021. Without weighting by state or district size — a small district’s decision counted as much as a large one’s — they then examined the respective racial and socioeconomic demographics of each area, their COVID positivity rates, the county-level vote share won by Hillary Clinton in the 2016 presidential election, as well as factors like local charter school enrollment and broadband access.

As in previous studies, they found that partisanship was an important determinant of whether schools opted for virtual learning to begin the 2020-21 school year. An increase in Clinton’s 2016 vote by 14 percent was associated with a 10.5 percentage point increase in the chance that local schools stayed remote. Whether teachers were allowed to collectively bargain was also linked to COVID decisions.

But a variety of demographic factors clearly play a role as well. Black, Hispanic, and low-income people living in a district were all associated with higher likelihoods that virtual learning continued last fall. Those trends dovetail with public opinion research released over the last year, which has that non-white parents are more concerned about the return to physical classrooms than their white counterparts.

Harris said that while 2016 voting patterns were the single strongest indicator of how districts acted, the totality of demographic evidence was equally compelling.

“You’ve really got one measure of political dynamics — two, if you include union strength — whereas demographics represents a set of factors,” said Harris. “Individually, they all look not as significant as that Democratic vote share, [but] collectively, they may be just as important.”

The effects of most district traits were substantially similar in both fall 2020 and spring 2021. Perhaps unsurprisingly, 2016 partisanship was somewhat less significant following the ultra-contentious 2020 election, even while it remained the most predictive single factor. The predictiveness of COVID positivity rates was also reduced, to the extent that it was no longer statistically significant.

To make matters more complex, Harris added, demographics and politics can’t be isolated from one another. Black and Hispanic voters, along with voters living in poverty, are more likely to vote Democratic; according to public health data they were also much more likely to find themselves at risk from the effects of COVID. All of it, Harris argued, made it necessary to avoid elevating one condition, such as politics, above others.

“Part of what we were trying to highlight was how connected those things were, and how it makes it pretty difficult when voting patterns are closely connected to demographics — race and income patterns — and those things, in turn, are closely correlated with COVID health risk. Whenever you have a bunch of interconnected factors like that, which are probably all playing some role, it becomes difficult to isolate one.”